Philosophy behind this machine migration is: ASM disk groups could be mounted to any machine, that see the underlying disk luns. And there is so much to do for OS admins then DBAs while migrating your database to new servers(machines).
Let's tell about our environment and requirements before dive into steps.
Both our database and grid infra versions are 11.2.0.2, running on IBM AIX 6.1 and migration/ upgrade from ibm p5 series machine to p6.
There are a single database and a RAC database in our RAC environment needed to migrate.
In our case, IP adress and VLAN are needed to be changed. That is why we need install grid and Oracle database software instead of only mounting u01 disk. Also we need to deal with scan name change side effects.
You could find the steps for changing database machines followingly:
1. Install grid infra, and Oracle database software and do not create data disk groups after then.
This step could be done at any time, even days before following steps.
2. Call application admins before starting.
[Steps 3,4,5 will be executed in old database node]
3. disable&stop listeners
--run below two lines in all nodes of cluster
srvctl stop listener -l listener
srvctl disable listener -l listener
--disable&stop scan_listeners in one node
srvctl disable scan_listener -i 1
srvctl disable scan_listener -i 2
srvctl disable scan_listener -i 3
srvctl stop scan_listener -i 1
srvctl stop scan_listener -i 2
srvctl stop scan_listener -i 3
4. Backup your archive logs to tape, in order to rollback the operation.
5. Stop database, asm, cluster.
crsctl stop database
crsctl stop cluster
crsctl stop crs
[Below steps will be executed in old database node]
6. Add database, instance to cluster
srvctl add database -d MYDB -c RAC -o /u01/app/oracle/product/11.2.0/dbhome_1
srvctl add database -d APEX -c SINGLE -x istanbul1 -o /u01/app/oracle/product/11.2.0/dbhome_1
srvctl add instance -d MYDB -i MYDB1 -n istanbul1
srvctl add instance -d MYDB -i MYDB2 -n istanbul2
srvctl add instance -d MYDB -i APEX -n istanbul1
8. Change asm search string as it was in older machines
alter system set asm_diskstring='/dev/ORACLE/*/*' scope=both;
9. Mount ASM diskgroups
alter diskgroup data1dg mount;
alter diskgroup data2dg mount;
10. Create init.ora files as in old servers.
/u01/app/oracle/product/11.2.0/dbhome_1/dbs/initMYDBDB1.ora and
/u01/app/oracle/product/11.2.0/dbhome_1/dbs/initMYDBDB2.ora
12. Create adump folder, give oracle user required OS rights
Check for audit_file_dest parameter in old system, create folder in all new nodes like below:
mkdir /u01/app/oracle/admin/MYDB/adump
chown oracle:oinstall mkdir /u01/app/oracle/admin/MYDB/adump
13. Start database [check alert.log while starting database]
srvctl start database -d MYDB
srvctl start database -d APEX
14. Add & start services to cluster [be sure you started MYDB database as in step 13]
srvctl add service -d MYDB -s MYSVC1 -r MYDB1 -a MYDB2
srvctl add service -d MYDB -s MYSVC2 -r MYDB1 -a MYDB2
srvctl start service -d MYDB -s MYSVC1
srvctl start service -d MYDB -s MYSVC2
15. Change remote_listener parameter in init.ora
ALTER SYSTEM SET remote_listener='mydb-scan.turkcell.tgc:1521' SCOPE=BOTH;
Subscribe to:
Post Comments (Atom)
Windows (powershell) counterparts of Linux commands
You may find Windows mostly powershell equivalent of frequently used Linux commands here. I will update this post, with newer ones by the ti...
3 comments:
One little addition: Before any change, be sure to find all application admins using your old database :) Thanks..
You are right... You cannot see database without application, and therefore application admins play great role during any change or improvement.
If we talk specifically for your case, problem came out from forwarded emails of mine, which were directly written to direct application responsibles.
Solution to this problem may be arranging or scheduling outlook appointments, in which scope, date of changes could be easily highlighted throughout the ecosystem of the service.
i have a question for below commands: how new server knows the disk group definitions like which asm disks included in which disk group ? because new server is installed freshly.
alter diskgroup data1dg mount;
alter diskgroup data2dg mount;
Post a Comment